Non-O157 Enterohemorrhagic Escherichia coli (EHEC) with AB5 Subtilase Cytotoxin (SubAB) Found in Commercial Ground Beef and Spinach Products


Introduction: Enterohemorrhagic Escherichia coli (EHEC) strains are a subset of Shiga toxin–producing E. coli (STEC) that may cause hemorrhagic colitis and in severe cases hemolytic uremic syndrome (HUS) and thrombotic thrombocytopenic purpura (TTP). EHECs are usually identified by the presence of two virulence genes, stx(encoding Shiga toxin) and eae (encoding intimin). AB5 subtilase cytotoxin (SubAB) was found in some non-O157 eae-negative STEC associated with HUS.

Purpose: This study surveyed the incidence of non-O157 STEC with eae and subAB in commercial ground beef products and spinach enrichments received for confirmatory testing in the laboratory.

Methods: A total of 1800 raw ground beef samples purchased prepackaged from various retail stores in the Seattle, WA area were analyzed from September, 2008 through June, 2009. Samples were enriched overnight in a selective media and tested by multiplex PCR for STEC detection. Enrichment cultures yielding EHEC presumptive positive bands (corresponding to stx, eae or subAB) were streaked onto MacConkey agar and incubated at 35°C overnight. Isolated colonies of pink color were analyzed by PCR to confirm the presence of the virulence factor genes and submitted for Pulsed Field Gel Electrophoresis (PFGE).

Results: Thirty two ground beef samples were positive for EHEC contamination. Of those 28 showed unique PFGE patterns. Twenty five strains (78.1%) had subAB with stx1, stx2. All of the strains with subAB were non-O157. Seven strains (21.9%) had eae with either stx1, or stx1 and stx2. No strains were positive for both subAB and eae. Additionally, from spinach samples sent for confirmation of EHEC contamination, five of seven EHEC-positive E. coli strains contained the subAB gene, and none of these were positive for eae.

Significance: Currently, E. coli O157:H7 is the only EHEC declared to be an adulterant in raw ground beef and is the only serotype for which USDA Food Safety and Inspection Service (FSIS) routinely tests. Certain strains of non-O157 STECs such as those containing the intimin or subtilase gene may be as virulent as E. coli O157 hence identifying these strains would be of value in improving the safety of food supply.


Shiga-like toxin (Stx)-containing Escherichia coli (STEC) infections from eating contaminated food are responsible for outbreaks of bloody diarrhea, hemorrhagic colitis and hemorrhagic uremic syndrome (HUS). E. coli has other toxins that also cause damage to cells, such as Intimin, Enterohemorrhagic E. coli hemolysin and Cytolethal distending toxin 5 (cdt5) (reviewed in (2)). Recently, a novel toxin designated Subtilase (SubAB) was identified from a 1998 HUS outbreak in Australia (10). Similarly to Stx, it is an AB toxin. The B portion binds to Neu5Gc, a glycoprotein on the cell surface, and the A portion is a protease that has recently been shown to cause cell death by compromising the function of the endoplasmic reticulum (6). This toxin was subsequently found in E. coli from food and fecal samples of cattle and many patients with diarrhea (4,8,9). These E coli were of various serotypes, but all were non-O157.

Currently, food, especially ground beef is routinely tested for STEC contamination. Commonly the tests are growth on sorbitol MacConkey agar (SMAC), which will differentiate E. coli O157 from all other serotypes, and PCR for stx and eae, the two toxins most frequently associated with outbreaks of HUS. However, given the presence of SubAB in food, as well as cattle and human feces, and its virulence, thought should be given to also testing for the presence of this toxin to ensure food safety. Here we demonstrate a multiplex PCR protocol that detects stx and subAB, and shows the presence of the genes for both of these toxins in not only ground beef, but also spinach and one sample of unmanipulated beef trim.


Between September, 2008 and June, 2009 we tested 1800 ground beef samples for the presence of stx and subAB genes in E. coli infecting these foods. Consumer-ready raw ground beef products were purchased prepackaged from various retail stores in the Seattle, WA area. The products were transferred to the laboratory at ambient temperature and processed immediately. Briefly, 200 g of each sample were aseptically transferred into a sterile Whirl-Pak sample bag (Nasco, Modesto, CA) and 800 ml of IEH Media (Acumedia, Neogen Corp, Lansing, MI.) was added. Meat samples were homogenized in an IUL Masticator (IUL, S.A., Spain)  for 60 s, and incubated at 35°C overnight for enrichment. Two microliters of enrichment culture were transferred to a tube containing commercially available multiplex PCR buffer for detection of stx1, stx2, eae and subAB genes (EC7 buffer, Molecular Epidemiology, Inc., Lake Forest Park, WA.) in a PCR assay. Amplification was performed using an Eppendorf Mastercycler (Eppendorf AG, Hamburg, Germany) according to the Technical Specifications for EC7, Molecular Epidemiology, Inc. Further PCR was run to identify those strains carrying the stx2d-activatable and cdt5 (5) genes. Amplicons were resolved by 2% agarose gel electrophoresis (Bio-Rad Laboratories sub-cell model 192, Hercules, CA.), and ethidium bromide stained gels were examined by UV illumination. E. coli were serotyped by PCR. Serotypes defined by PCR were O26, O103, O111, O113, O121 and O145. Those strains that did not yield PCR signals were serotyped using conventional methods by the National Microbiology Laboratory, Winnipeg, MB, Canada. Isolates were analyzed by PFGE according to CDC Pulse Net protocol for molecular subtyping of E.coli ecoli_salmonella_shigella_protocols.pdf with slight modifications (Specifications for PFGE, Molecular Epidemiology, Inc., Lake Forest Park, WA). PFGE gel images were analyzed with BioNumerics software (Applied Maths, Inc., Austin, Texas). Cluster analysis was performed using the Dice similarity coefficient and UPGMA (Unweighted Pair Group Method with Arithmetic mean) algorithm to construct a dendrogram.


Of the 32 ground beef samples that were positive for E. coli, 28 had unique PFGE patterns. subAB was detected in 25 (78.1%) samples. Five of seven E. coli-infected spinach samples and one beef trim also tested positive for subAB (fig. 1). All but one of the food samples carrying the subAB gene had stx2, seven of the ground beef samples also had the stx1 gene and one had only stx1 (see table 1). All but two of the subAB-negative samples contained only stx1. All ground beef samples positive for the subAB gene were negative for the eae gene (table 1). Serotypes of the E. coli found in the food were varied, but O26 was found only in the subAB negative samples. Additionally, cdt5 was found in subAB positive strains of O113 and O91 only.


Figure 1 – Comparison of the number E. coli-containing food samples that were positive versus negative for the subAB gene.


Figure 2 – Dendromgram of PFGE results showing the banding pattern of 32 ground beef samples and three subAB/stx2(d-activatable) E. coli isolates from ATCC for comparison.




The SubAB toxin was first identified in a HUS outbreak in Australia, and has since shown to be highly toxic for Vero cells and lethal in a mouse model (9,13). It inactivates the ER chaperone BiP, causing cell death (4). However, current food testing does not include identification of this toxin. SubAB was identified and characterized relatively recently, and questions about its associations with other toxins and serotypes and prevalence are just now starting to be answered as more strains of E. coli are tested for the presence of SubAB. Paton and Paton (8) devised a multiplex PCR system to detect the genes of several E. coli toxins, among them SubAB and Stx1 and 2, in strains from human fecal samples. We used a modified version of this multiplex PCR for food samples. Similarly to them, we found subAB in in strains that contained primarily stx2, all were eae negative, and were of several serotypes in addition to O113 that was found in the Australian outbreak. subAB was present in 78% of the ground beef samples that we tested, compared to 11% of the strains from human samples previously tested (8). subAB was also found in 44% of E. coli strains from cattle that were tested in Brazil, of these 57% contained stx2 only, and again they were of various serotypes but all non O157 (4). Four of the subABpositive samples also contained the gene for Cdt5, a toxin previously associated with diarrhea in human patients (1,5,14).

We found E. coli containing subAB not only in beef samples tested but also in spinach, indicating likely waterborne transmission. It was also present in E. coli cell lines obtained from ATCC (see dendrogram), previously not tested for subAB. Obviously, many strains of E. coli contain the gene for SubAB toxin because we identified unique PFGE patterns and many different serotypes, although none were O157. E. coli O157 strains are responsible for the majority of HUS outbreaks and serious diarrheal disease, but a negative SMAC test alone cannot rule out pathogenic bacteria contamination. Strikingly, one of our subAB positive meat samples was stx-negative and not of the O157 serotype and would have been completely missed with conventional testing methods including PCR for the stx gene.

SubAB may contribute to disease severity or act alone to confer illness. In mice it causes damage to several organs including the brain, kidneys, liver and spleen (13). It has been found in E. coli strains infecting people and now we have shown the presence of this toxin in food samples, so identification of this toxin should be added to the current standard procedure for testing of food.


  1. Bielaszewska, M., and H. Karch. 2005. Consequences of enterohaemorrhagic Escherichia coli infection for the vascular endothelium. Thromb Haemost 94:312-8.
  2. Boerlin, P., S. A. McEwen, F. Boerlin-Petzold, J. B. Wilson, R. P. Johnson, and C. L. Gyles. 1999. Associations between virulence factors of Shiga toxin-producing Escherichia coli and disease in humans. J Clin Microbiol 37:497-503.
  3. Cergole-Novella, M. C., L. S. Nishimura, L. F. Dos Santos, K. Irino, T. M. Vaz, A. M. Bergamini, and B. E. Guth. 2007. Distribution of virulence profiles related to new toxins and putative adhesins in Shiga toxin-producing Escherichia coli isolated from diverse sources in Brazil. FEMS Microbiol Lett 274:329-34.
  4. Paton, A. W., T. Beddoe, C. M. Thorpe, J. C. Whisstock, M. C. Wilce, J. Rossjohn, U. M. Talbot, and J. C. Paton. 2006. AB5 subtilase cytotoxin inactivates the endoplasmic reticulum chaperone BiP. Nature 443:548-52.
  5. Paton, A. W., P. A. Manning, M. C. Woodrow, and J. C. Paton. 1998. Translocated intimin receptors (Tir) of Shiga-toxigenic Escherichia coli isolates belonging to serogroups O26, O111, and O157 react with sera from patients with hemolytic-uremic syndrome and exhibit marked sequence heterogeneity. Infect Immun 66:5580 6.
  6. Escherichia coli strains producing the novel subtilase cytotoxin. J Clin Microbiol 43:2944-7.
  7. Paton, A. W., P. Srimanote, U. M. Talbot, H. Wang, and J. C. Paton. 2004. A new family of potent AB(5) cytotoxins produced by Shiga toxigenic Escherichia coli. J Exp Med 200:35-46.
  8. Paton, A. W., M. C. Woodrow, R. M. Doyle, J. A. Lanser, and J. C. Paton. 1999. Molecular characterization of a Shiga toxigenic Escherichia coli O113:H21 strain lacking eae responsible for a cluster of cases of hemolytic-uremic syndrome. J Clin Microbiol 37:3357-61.
  9. Paton, J. C., and A. W. Paton. 1998. Pathogenesis and diagnosis of Shiga toxinproducing Escherichia coli infections. Clin Microbiol Rev 11:450-79.
  10. Takeda, Y., H. Kurazono, and S. Yamasaki. 1993. Vero toxins (Shiga-like toxins) produced by enterohemorrhagic Escherichia coli (verocytotoxin-producing E. coli). Microbiol Immunol 37:591-9. Wang, H., J. C. Paton, and A. W. Paton. 2007. Pathologic changes in mice induced by subtilase cytotoxin, a potent new Escherichia coli AB5 toxin that targets the endoplasmic reticulum. J Infect Dis 196:1093-101.

- Viktoriya Beskhlebnaya, Kay Greeson, Songhai Shen, Ramon Aboytes and Mansour Samadpour